diff --git a/m-bitset.h b/m-bitset.h
index fbbc46cf..bb82d353 100644
--- a/m-bitset.h
+++ b/m-bitset.h
@@ -786,6 +786,36 @@ bitset_clz(const bitset_t set)
   return s;
 }
 
+// For GCC or CLANG or ICC
+#if defined(__GNUC__)
+static inline size_t m_b1tset_popcount64(bitset_limb_ct limb)
+{
+  return (size_t) __builtin_popcountll(limb);
+}
+#else
+// MSVC __popcnt64 may not exist on the target architecture (no emulation layer)
+// Use emulation layer: https://en.wikipedia.org/wiki/Hamming_weight
+static inline size_t m_b1tset_popcount64(bitset_limb_ct limb)
+{
+  limb = limb - ((limb >> 1) & 0x5555555555555555ULL);
+  limb = (limb & 0x3333333333333333ULL) + ((limb >> 2) & 0x3333333333333333ULL);
+  limb = (limb + (limb >> 4)) & 0x0f0f0f0f0f0f0f0fULL;
+  return (limb * 0x0101010101010101ULL) >> 56;
+}
+#endif
+
+/* Count the number of 1 */
+static inline size_t
+bitset_popcount(const bitset_t set)
+{
+  M_B1TSET_CONTRACT(set);
+  size_t s = 0;
+  size_t n = (set->size + M_B1TSET_LIMB_BIT - 1) / M_B1TSET_LIMB_BIT;
+  for(size_t i = 0 ; i < n; i++)
+    s += m_b1tset_popcount64(set->ptr[i]);
+  return s;
+}
+
 /* Oplist for a bitset */
 #define BITSET_OPLIST                                                         \
   (INIT(bitset_init)                                                          \
diff --git a/tests/test-mbitset.c b/tests/test-mbitset.c
index d3488dfc..c65f43d3 100644
--- a/tests/test-mbitset.c
+++ b/tests/test-mbitset.c
@@ -366,6 +366,17 @@ static void test_clz(void)
     bitset_set_at(s1, 0, false);
     assert (bitset_clz(s1) == 64+620+1);
   }
+
+  M_LET(s1, bitset_t) {
+    bitset_set_str(s1, "[00000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000]");
+    assert(bitset_popcount(s1) == 0);
+    bitset_set_str(s1, "[10000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000]");
+    assert(bitset_popcount(s1) == 1);
+    bitset_set_str(s1, "[00000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000001]");
+    assert(bitset_popcount(s1) == 1);
+    bitset_set_str(s1, "[10000011100000110010101010110000011111111111111110000001110001110001110001110000]");
+    assert(bitset_popcount(s1) == 40);
+  }
 }
 
 int main(void)